Research On Magnetic Compensation Algorithm Based On Optimized BP Neural Network

Aeromagnetic compensation is an indispensable technology in the aviation magnetic detection process.Current aeromagnetic compensation methods rely mainly on the Tolles-Lawson model.This model divides the vehicle interference field into three types of interference sources: constant field,induction field and eddy current field.It combines the attitude information of the vehicle and expresses it in the form of linear equations.Calibration is accomplished by clearing the model coefficients and compensation is achieved.However,there are still many problems in the implementation of this method.For example,the accuracy of attitude data is limited,and the model lacks low-frequency disturbance description items.These problems make the magnetic compensation algorithm based on Tolles-Lawson model difficult to obtain higher compensation accuracy.With the rapid progress in the measurement accuracy of magnetometers,the accuracy of compensation has become a major bottleneck in the accuracy of aeromagnetic surveys.Based on the above issues,this paper will start with the aeromagnetic compensation model and attitude data quality,and study the new aeromagnetic compensation technology.For the aeromagnetic compensation model,a new BP neural network model is established based on the traditional Tolles-Lawson model,and the weight initialization method is designed.For the attitude data quality,this paper combines the three-component fluxgate magnetometer.Data and attitude data improve data quality.In summary,the research content of this paper mainly includes the following three aspects:First,establish a BP neural network magnetic interference compensation model.This paper first analyzes the advantages and disadvantages of the traditional TollesLawson aeromagnetic compensation model,and discusses the feasibility of using BP neural network to perform magnetic interference compensation.Next,this paper designs a multi-group network structure and tests it.Data was verified.Experiments show that the compensation model based on neural network can improve the compensation effect to some extent.Second,the method of solving the aeromagnetic interference model for genetic algorithm is proposed.This chapter starts from the pre-training of BP neural network weights,and discusses the application of genetic algorithms in voyage disturbance compensation.Firstly,the genetic algorithm is used to initialize the BP neural network weights;then the genetic algorithm is extended to the traditional Tolles-Lawson model coefficients to improve the accuracy of aeromagnetic compensation.Thirdly,an attitude data quality enhancement algorithm based on data fusion is proposed.This chapter first analyzes the respective errors and compensation results of the three-component fluxgate and the attitude instrument,and based on this,designs a fusion algorithm to obtain higher-precision attitude data,thus improving the accuracy of compensation.